TABLE 4.

Steps of thematic analysis.

Step	Details
Familiarisation	Repeat listening to audiofiles to facilitate data immersion; repeat reading of transcripts while noting points of interest, coding ideas, reactions to the data, ideas for new interview questions, repetitive or novel issues, and preliminary thematic ideas.
Inductive coding	Systematic reading of transcripts to identify units of text relevant to the research questions, and to assign each chunk of text a label.
Constructing themes	Analysis of the content of each code and relationship between codes, grouping codes into broader themes. e.g., family statement, family witness, and questioning witnesses, were grouped into a theme provisionally labelled “active participation” because these codes reflected the family’s experience of active contribution to the investigation. Eleven preliminary themes were identified in relation to family experiences of the coronial process and best practice.
Naming themes	Named themes and subthemes to convey the underlying meaning of the theme. For example, “active participation” was renamed “opportunity to be heard” to preserve a phrase used repeatedly by participants.
Themes revision	Explored the consistency of data coded within each theme, differentiation between themes, and the relationships between themes. Themes were identified as salient or predominant if they intersected with a number of other themes, if several codes clustered within that theme, or if participants labelled the concept as important or as having a significant impact on well-being.
Establishing trustworthiness	The reliability of findings was established by: (a) allowing participants to check and edit transcripts, (b) reflexive journaling of research steps, (c) development of a codebook to enhance replicability of decisions, (d) discussion and review of analytic steps with co-authors experienced in qualitative research, (e) mapping thematic concepts, and (f) independently comparing raw data (transcripts) with final themes to validate interpretations (Nowell et al., 2017).